Linear Software Models Are Theoretical Standards of Modularity

Abstract

Modularity is essential to obtain well-composed software systems from COTS (Commercial Off-The-Shelf) components. But COTS components do not necessarily match the modules of the designed software system. A clear-cut procedure is needed for the choice of the necessary and sufficient components providing the required functionalities. Linear Software Models are rigorous theoretical standards of modularity. These models are proposed as formal criteria for well-composed software systems. The paper lays down theoretical foundations - upon linear independence and reducible matrix concepts - providing precise meanings to familiar software concepts, such as coupling and the single responsibility theorem. The theory uses a Modularity Matrix - linking linearly independent software structors to composable software functionals. The theory has been tested by applying it to small canonical systems and to large software systems found in the literature. © Springer-Verlag Berlin Heidelberg 2013.